Manuscript Details Manuscript number JAPG_2019_1240 Title Reducing deforestation through value chain interventions in countries emerging from conflict: the case of the Colombian cocoa sector Article type Full Length Article Abstract Sustainability commitments by private sector actors are emerging as promising interventions to help reduce global deforestation. Much attention is placed on the forest conservation impact of these interventions in areas where commodity production constitutes a main driver of deforestation. It is however less clearly understood what role they could play in areas where the production of commodities is not evidently leading to the loss of forest, and how they could contribute to other objectives including sustainable rural development and peacebuilding. In this paper, we examine the potential of the cocoa value chain in Colombia in achieving deforestation reduction and peacebuilding simultaneously, as aimed by the country’s Cocoa, Forests and Peace Initiative. Results from correlations and spatially explicit analyses show that regardless of its widespread production across Colombia, cocoa is not an important driver of deforestation. This suggest that efforts to end deforestation in the Colombian cocoa sector emerged following global trends, and not because of an evident link between cocoa production and deforestation. Furthermore, results from spatial clustering analyses highlight areas where different types of value chain interventions may be appropriate to parallel forest conservation and peacebuilding, while interviews with key actors in the cacao sector provide clues as to how these interventions should be developed and implemented. Specifically, our results show that narratives around approaches to achieve zero-deforestation from agricultural commodities should (1) be adjusted to local contexts, (2) incorporate location-specific development needs, (3) complement existing rural development efforts, (4) enhance collaboration among actors that operate both within and beyond the value chain, and (5) apply high-resolution data to assess deforestation-commodity relations and verify zero-deforestation commitments. These considerations are particularly relevant in contexts where commodity production is not evidently leading to deforestation, as in the case of cocoa production in Colombia. Keywords Zero-deforestation, agricultural commodity, value chain, peacebuilding, land use change Corresponding Author Augusto Castro-Nunez Corresponding Author's University of Copenhague Institution Order of Authors Augusto Castro-Nunez, Andres Charry Camacho, Fabio Castro, Janelle Sylvester, Vincent Bax Submission Files Included in this PDF File Name [File Type] Letter.docx [Cover Letter] Tittle page_cacao, forests and peace.docx [Title Page (with Author Details)] Castro- Nunez et al. Reducing deforestation in the cacao sector_Colombia.docx [Manuscript (without Author Details)] To view all the submission files, including those not included in the PDF, click on the manuscript title on your EVISE Homepage, then click 'Download zip file'. Research Data Related to this Submission There are no linked research data sets for this submission. The following reason is given: The authors do not have permission to share data Dr. N. Hoalst-Pullen Date: 17-12-2019 Editor in Chief: Applied geography Dear Dr. Hoalst-Pullen We are pleased to submit our manuscript entitled “Reducing deforestation through value chain interventions in countries emerging from conflict: the case of the Colombian cocoa sector” for your consideration for publication as a research article in Applied Geography. In this paper, we examine the potential of the cocoa value chain in Colombia in achieving deforestation reduction and peacebuilding simultaneously, as aimed by the country’s Cocoa, Forests and Peace Initiative. Results from correlations, spatially explicit analyses and interviews with key actors in the value chain provide insights on how these interventions should be developed and implemented. This manuscript describes original work and is not under consideration for publication elsewhere. All authors have approved the manuscript and agree with the submission. Thank you for receiving our manuscript and considering it for review. We look forward to hearing from you at your earliest convenience. Sincerely, Augusto Castro-Nunez (on behalf of all authors) International Center for Tropical Agriculture Km 17, Recta Cali-Palmira Apartado Aéreo 6713 Cali, Colombia augusto.castro@cgiar.org Reducing deforestation through value chain interventions in countries emerging from conflict: the case of the Colombian cocoa sector Authors Augusto Castro-Nunez1,*, Andres Charry1, Fabio Castro1, Janelle Sylvester1, Vincent Bax2 1 International Center for Tropical Agriculture (CIAT), Km 17 recta Cali-Palmira, Cali 763537, Colombia; 2 Centre for Interdisciplinary Science and Society Studies, Universidad de Ciencias y Humanidades, Av. Universitaria 5175, Los Olivos, Lima, Peru *corresponding author (augusto.castro@cgiar.org) Acknowledgements This research was developed as part of the project 18_III_106_COL_A_Sustainable productive strategies. This project is part of the International Climate Initiative (IKI). The Federal Ministry for the Environment, Nature Conservation and Nuclear Safety (BMU) supports this initiative on the basis of a decision adopted by the German Bundestag. 1 Reducing deforestation through value chain interventions in countries 2 emerging from conflict: the case of the Colombian cocoa sector 3 4 Abstract 5 Sustainability commitments by private sector actors are emerging as promising interventions to 6 help reduce global deforestation. Much attention is placed on the forest conservation impact of 7 these interventions in areas where commodity production constitutes a main driver of deforestation. 8 It is however less clearly understood what role they could play in areas where the production of 9 commodities is not evidently leading to the loss of forest, and how they could contribute to other 10 objectives including sustainable rural development and peacebuilding. In this paper, we examine 11 the potential of the cocoa value chain in Colombia in achieving deforestation reduction and 12 peacebuilding simultaneously, as aimed by the country’s Cocoa, Forests and Peace Initiative. 13 Results from correlations and spatially explicit analyses show that regardless of its widespread 14 production across Colombia, cocoa is not an important driver of deforestation. This suggest that 15 efforts to end deforestation in the Colombian cocoa sector emerged following global trends, and 16 not because of an evident link between cocoa production and deforestation. Furthermore, results 17 from spatial clustering analyses highlight areas where different types of value chain interventions 18 may be appropriate to parallel forest conservation and peacebuilding, while interviews with key 19 actors in the cacao sector provide clues as to how these interventions should be developed and 20 implemented. Specifically, our results show that narratives around approaches to achieve zero- 21 deforestation from agricultural commodities should (1) be adjusted to local contexts, (2) 22 incorporate location-specific development needs, (3) complement existing rural development 23 efforts, (4) enhance collaboration among actors that operate both within and beyond the value 24 chain, and (5) apply high-resolution data to assess deforestation-commodity relations and verify 25 zero-deforestation commitments. These considerations are particularly relevant in contexts where 26 commodity production is not evidently leading to deforestation, as in the case of cocoa production 27 in Colombia. 28 Keywords 29 Zero-deforestation, agricultural commodity, value chain, peacebuilding, land use change 1 30 1. Introduction 31 Global clamor over the need to reduce deforestation linked to agricultural production in order to 32 lower carbon emissions and curb the loss of biodiversity is increasing. Commitments on 33 sustainability by private sector actors are emerging as promising interventions to help reduce global 34 deforestation (Lambin et al., 2018). Hundreds of corporations have pledged to enhance 35 transparency and accountability in their supply chains as a means to achieve zero-deforestation. 36 However, the impact of such commitments on reducing deforestation has been limited (Garrett et 37 al., 2019). Greater impacts may be accomplished through the implementation of value chain 38 interventions (VCI), here defined as actions directed at segments of a value chain, or along its entire 39 length, to achieve certain environmental, social or economic development goals (Sola et al., 2017; 40 Zuberi, Mehmood & Gazdar, 2016). 41 Zero-deforestation VCI provide an opportunity to put zero-deforestation commitments into action. 42 However, such interventions are facing various challenges in reaching desired outcomes (Garrett 43 et al., 2019). In fact, before gaining prominence as a tool to achieve zero-deforestation, VCI were 44 promoted as a means to deliver sustainable development, including conflict resolution, poverty 45 reduction, rural development, gender inclusion, improved nutrition, food security and forest 46 conservation (Bolwig, Ponte, du Toit, Riisgaard & Halberg, 2008; Devaux, Torero, Donovan & 47 Horton, 2018; Maestre, Poole & Henson, 2017; Seville, Buxton & Vorley, 2011; Tallontire & 48 Vorley, 2005; Zuberi et al., 2016). Nevertheless, the impact of such interventions on sustainable 49 development remains a topic of debate (Kidoido & Child, 2014). This is partly because value chains 50 are complex, multi-layered in nature, highly diverse, dynamic, and time and context-specific 51 (Devaux et al., 2018; Kidoido et al., 2014; Reardon et al., 2019; Ton, Vellema & de Ruyter de 52 Wildt, 2011). Furthermore, some authors argue that to achieve sustainable outcomes, VCI alone 53 are insufficient, and they need to be implemented concertedly with other sustainability approaches, 54 engage stakeholders along the entire value chain, and address multiple factors and interactions 55 (Devaux et al., 2018; Seville et al., 2011). 56 Emerging literature on the topic mainly focuses on assessing corporate supply-chain commitments 57 in contexts where there is a clear link between an agricultural commodity and deforestation 58 (Gardner et al., 2019; Garrett et al., 2019; Lambin et al., 2018). However, contexts where this link 59 is weak or unapparent are often disregarded. Zero-deforestation initiatives were initially developed 60 to reduce forest loss in countries where globally traded commodities are the main drivers of 61 deforestation, such as in Brazil, Indonesia and Malaysia (Boucher & Elias, 2013; Gibbs et al., 2015; 62 Henders, Persson & Kastner, 2015). These initiatives were built upon “name and shame” 63 campaigns that have led to recent trends of incorporating social and environmental concerns into 64 corporations’ supply chains (Vurro, Russo & Perrini, 2009) and implementing governance models 65 that encompass extensive collaboration with all stakeholders involved in the value chain (Jiang, 66 2009). Arguments for such interventions are supported by evidence indicating that the production 67 of globally traded agricultural commodities – such as palm oil, soy, beef, coffee and cocoa – is an 68 overwhelming cause of tropical deforestation (McCarthy & Tacconi, 2011). 69 On the other hand, there is no clear understanding of the role of such interventions in reducing or 70 preventing deforestation where forest cover changes are tied to interlinkages that are more 71 complex. For instance, the expansion of commodities on previously cleared land may have limited 72 impacts on forest cover, and could even contribute to reforestation in the case of tree crops such as 73 cocoa (Schroth, Garcia, Griscom, Teixeira & Barros, 2016). Similarly, it is not clear how these 74 interventions should be developed and implemented in such contexts, particularly because on-the- 2 75 ground implementation would naturally vary between countries where deforestation is driven by 76 agricultural commodities and those where it is not (McCarthy et al., 2011). Implementation should 77 also vary between cases that involve global brands that dominate the market and small producers 78 located in isolated regions, such as regions emerging from armed conflict (Perez-Aleman & 79 Sandilands, 2008; Reed & Reed, 2009; Rein & Stott, 2009). For instance, there is no doubt that in 80 some parts of the globe, cocoa farming has led to deforestation; the sector, hence, has been the 81 subject of sharp criticism. This is the case for countries where cocoa has been promoted as an 82 economic alternative in post-conflict settings, such as Ghana (Deans, Ros-Tonen & Derkyi, 2018). 83 It is not clear, however, as to what extent cocoa is causing deforestation in other countries, such as 84 Colombia, where i) complex interlinkages between coca leaf production, cattle pastures and land 85 grabbing drive deforestation (Castro-Nunez, Mertz, Buritica, Sosa & Lee, 2017), ii) the cultivation 86 of cocoa has been promoted as an alternative to illegal crop production (Charry, Castro-Llanos & 87 Castro-Nunez, 2019), and iii) most of the cocoa production is traded nationally (Abbott et al., 88 2018)). 89 In this paper, we contribute to the understanding of the role of VCI in achieving zero-deforestation 90 in areas where the link between deforestation and commodity production is not evident. It does so 91 via quantitative analyses and interviews with key stakeholders in the cocoa sector in Colombia. We 92 use Colombia, a country emerging from armed conflict, as a case study; because, despite a lack of 93 evidence in the literature that cocoa is causing significant deforestation, Colombia’s government 94 has joined global efforts to achieve deforestation-free cocoa production, which is being carried out 95 under the country’s Cocoa, Forests and Peace Initiative (Minambiente, 2018). We first perform 96 correlations and spatially explicit analyses to explore to what degree deforestation is associated 97 with cocoa production in Colombia, and examine how locations with similar cocoa, forest and 98 conflict characteristics are spatially distributed. We then use semi-structured interviews with key 99 stakeholders to better understand their viewpoint about the potential role of the cocoa value chain 100 as a tool for forest conservation and peacebuilding and identify opportunities and barriers in 101 delivering both forest conservation and long-lasting peace, as aimed by the Cocoa, Forests and 102 Peace Initiative. After this introductory section, the methods are described. Subsequently, results 103 from Spearman correlation analysis, Local Indicators of Spatial Association (LISA) analysis, 104 Hierarchical Cluster Analysis and interviews are presented. We then discuss the implications of 105 our results for on-the-ground implementation of zero-deforestation VCI. 106 107 2. Methods 108 2.1 Cocoa, forests and peace in Colombia 109 Since the beginning of the peace negotiations between the Colombian government and the 110 Revolutionary Armed Forces of Colombia (FARC), the country has been experiencing changes in 111 multiple dimensions, which has brought about new environmental, social and political challenges 112 (Eufemia et al., 2019). Approximately 52% of Colombia’s 114.2 million hectares of land are 113 covered with natural forests. Around 60% of its natural forests are found in the Amazon region, 114 while 17% and 9% are found in the Colombian Andes and Pacific region, respectively. According 115 to the Colombian Institute of Hydrology, Meteorology and Environmental Studies (IDEAM), more 116 than 5.6 million hectares of forests were lost between 1990 and 2010, with an average annual 117 deforestation rate of 0.42% between 1990 and 2000, a rate of 0.52% between 2000 and 2005, and 118 a rate of 0.47% between 2005 and 2010. Lower deforestation was observed between 2010 and 119 2013, at an average rate of 0.28% per year (IDEAM, 2018). 3 120 Deforestation has been particularly severe in areas affected by the armed conflict and illicit crop 121 production (Charry et al., 2019). In these areas, cocoa cultivation has been promoted by the 122 Government of Colombia (GoC) and international cooperation agencies as a productive alternative 123 to illicit crops for several decades. In light of recent global trends to achieve zero-deforestation in 124 agricultural value chains, several actors have highlighted existing opportunities to produce cocoa 125 with zero-deforestation in areas prioritized for peacebuilding and rural development efforts. These 126 areas include municipalities defined by the GoC as Areas Most Affected by Armed Conflict 127 (ZOMAC) and prioritized for Development Programs with Territorial Approach (PDET). For 128 example, in July 2018, the Cocoa, Forests and Peace Initiative was signed by the GoC, producer 129 associations, the National Federation of Cocoa Producers (FEDECACAO), industry 130 representatives and national and international civil society organizations. The signatories have 131 agreed to work together to end deforestation and promote forest protection and restoration through 132 the Framework Agreement for Joint Action, which is structured around the following three priority 133 areas: (1) forest protection and restoration; (2) sustainable cocoa production and livelihood security 134 of farmers; and (3) community participation and social inclusion. 135 Unlike other producing countries, most of Colombia’s cocoa production is used to meet domestic 136 demand. Production occurs mostly within the departments of Santander and Nariño in the north 137 and south respectively, where many conflict zones and areas emerging from conflict are located. 138 Production occurs on a smaller scale in the departments of Antioquia, Arauca, Tolima and Norte 139 de Santander. Cocoa is mainly produced by small producers (about 90%), who typically plant 140 around 3 hectares of cocoa (Abbott et al., 2018). Figure 1 illustrates the average cocoa crop area 141 per department between 2007 and 2017. 4 142 143 Figure 1. Average crop area of cocoa cultivation at the department level between 2007 and 2017 as reported by the 144 Colombian Ministry of Agriculture and Rural Development. 145 146 2.2 Correlations and spatially explicit analyses 147 We first examined to what degree deforestation is associated with cocoa production in Colombia 148 and how areas with similar cocoa, forests and conflict characteristics are spatially distributed. 149 Analyses were limited to cocoa-producing municipalities in Colombia (n = 529), taking the 150 municipality as the unit of analysis. This study utilized official data on cocoa from the Ministry of 151 Agriculture and Rural Development (MADR, 2018) and data on forest cover from IDEAM 152 (IDEAM, 2018), (Table 1). Selection of data sources depended on data availability at the municipal 153 level. 154 Spearman’s rank correlation coefficients (rs) were calculated to identify correlations among the 155 following 6 variables related to forest cover, cocoa production, and the armed conflict: (1) forest 156 area; (2) change in forest cover; (3) cocoa area; (4) cocoa production; (5) cocoa yields; and (6) 157 armed conflict index. In addition, we examined local spatial associations between “change in forest 158 cover” and “cocoa area” by computing bivariate local Moran’s I values, also known as Local 159 Indicators of Spatial Association (LISA) (Anselin, 1995). The municipalities were then clustered 160 using Hierarchical Cluster Analysis (Euclidean distances and Ward’s method), which included five 161 variables: (1) forest area; (2) change in forest cover; (3) cocoa area; (4) cocoa yields; and (5) land 5 162 suitable for cocoa cultivation. The clusters are composed of municipalities that are similar with 163 respect to the cocoa production and forest cover variables included in the analysis. The location of 164 the clusters were mapped to examine how they are spatially distributed. Then, the number of 165 municipalities defined by the GoC as Areas Most Affected by Armed Conflict (ZOMAC) and 166 prioritized for Development Programs with Territorial Approach (PDET) within each cluster were 167 identified. The Hierarchical Cluster Analysis was conducted using the Stats package in R v2.9.2 168 (R Core Team) and bivariate local Moran’s I values were estimated using the software GeoDa 169 v1.6.7.9 (Anselin, Syabri & Kho, 2006). 170 Table 1. Variables used to examine the relationship among cocoa cultivation, changes in forest cover and the armed 171 conflict in Colombia. Variable Period Source * Forest area (%) 2017 IDEAM Change in forest cover (%) 2005 - 2017 IDEAM Average cocoa area (ha) 2007 - 2017 MADR Average cocoa production (ton) 2007 – 2017 MADR Average cocoa yields (ton/ha) 2007 – 2017 MADR Land suitable for cocoa cultivation (%) 2017 UPRA Armed conflict index1 2016 DNP ZOMAC municipalities2 2016 DNP PDET municipalities3 2017 DNP 172 1 Index ranging from 0 to 1, with values 0 indicating no conflict and 1 indicating high conflict 173 2 Municipalities defined as Areas Most Affected by Armed Conflict 174 3 Municipalities prioritized for Development Programs with Territorial Approach 175 * Data sources: Instituto de Hidrología, Meteorología y Estudios Ambientales Institute of Hydrology, Meteorology and 176 Environmental Studies (IDEAM); Ministry of Agriculture and Rural Development (MADR); Rural Agricultural 177 Planning Unit (UPRA); National Planning Department (DNP) 178 179 2.3 Semi-structured interviews 180 The quantitative analyses described above were coupled with 30 semi-structured interviews 181 conducted with key actors in the cocoa value chain. The interviews were conducted from December 182 2018 to January 2019 and focused on: (1) the role of cocoa in forest conservation, restoration and 183 peace processes, and (2) opportunities and limitations for the development of the national cocoa 184 value chain. Respondents were selected from all three levels of the value chain (i.e. micro, meso 185 and macro; according to the classification proposed by Jäger, Jiménez & Amaya (2013)). Several 186 of the respondents fulfilled more than one role within the value chain. The most represented role 187 was related to the processing of cocoa into chocolate or other products, with eight interviewees 188 assuming this role. Other roles represented by respondents included artisan chocolatiers, 189 representatives of the Bean to Bar sector, large industrial companies, academia, producers, union 190 representatives, policymakers and other government employees (from ministries and other public 191 entities), representatives from international cooperation agencies, and suppliers of plant material 192 and other business services. 6 193 3. Results 194 3.1 Correlations among cocoa, forests and conflict in Colombia 195 A Spearman correlation matrix was computed based on the 529 cocoa-producing municipalities in 196 Colombia (Table SM1 in the supplementary material). The results show that mainly weak 197 correlations (rs <0.30, p < 0.05) exist among the forest, cocoa and conflict variables included in the 198 analysis. However, a positive, moderate correlation was found between “forest area” and “armed 199 conflict index” (rs = 0.46, p < 0.05), and a strong positive correlation between “cocoa production” 200 and “cocoa area” (rs = 0.76, p < 0.05). 201 202 3.2 Local spatial associations between deforestation and cocoa production (LISA) 203 The distribution of bivariate Moran’s I values sheds light on local patterns of spatial associations 204 between “change in forest cover” and “cocoa area” at municipality level (Figure 2). More 205 specifically, the results point to statistically significant spatial associations (p < 0.05) between 206 deforestation in a given municipality and cocoa area in a neighboring municipality, for 118 out of 207 the 529 municipalities included in the analysis. 208 In 11 out of the 118 municipalities, high deforestation pressure is spatially associated with high 209 cocoa production within neighboring municipalities (High-High associations; highlighted in red in 210 Figure 2). This suggests that only in and around these 11 municipalities, the production of cocoa 211 could potentially be a driver of forest loss. However, further analysis would be needed to attribute 212 causal relationships. Nine out of eleven municipalities are located in the departments of Nariño and 213 Santander, where currently most of Colombia’s cocoa production is taking place. One municipality 214 in Santander has been defined as ZOMAC. In addition, five municipalities in Nariño have been 215 defined as ZOMAC, of which two also have been prioritized for PDET. 216 In 49 municipalities, high deforestation pressure is spatially associated with low cocoa production 217 within neighboring municipalities (High-Low associations; highlighted in orange in Figure 2). This 218 implies that in these areas, activities other than cocoa cultivation seem to be driving deforestation. 219 More than half of the municipalities characterized by High-Low associations are located in the 220 departments of Casanare, Meta and Valle del Cauca. Five municipalities have been prioritized for 221 PDET, while 25 have been defined as ZOMAC. 222 In 24 municipalities, low deforestation pressure is spatially associated with high cocoa production 223 in neighboring municipalities (Low-High associations; highlighted in light blue in Figure 2), which 224 points to extensive cocoa production activities with, nonetheless, limited impacts on forest cover. 225 These municipalities are mainly located in the departments of Boyacá and Santander. Two of them 226 (both located in Santander) have been defined as ZOMAC, while none have been prioritized for 227 PDET. 228 In 34 municipalities, low deforestation pressure is spatially associated with low cocoa production 229 within neighboring municipalities (Low-Low associations; highlighted in dark blue in Figure 2). 230 This points to limited cocoa production activities that, in turn, generate a low impact on forest 231 cover. Almost 60% of these municipalities have been prioritized for PDET and are mainly located 232 in the Amazon region, within the departments of Caquetá (6 municipalities), Meta (5 233 municipalities), Putumayo (5 municipalities) and Guaviare (3 municipalities). More than 80% of 234 the municipalities characterized by Low-Low associations have been defined as ZOMAC. 7 235 236 Figure 2. Moran cluster map of the local spatial associations between “change in forest cover” and “cocoa area” at 237 municipality level. High-High associations point to high deforestation surrounded by high cocoa area (red); High-Low 238 associations point to high deforestation surrounded by low cocoa area (orange); Low-High associations point to low 239 deforestation surrounded by high cocoa area (light blue); Low-Low associations point to low deforestation surrounded 240 by low cocoa area (dark blue). Municipalities with non-significant local Moran’s I values (p > 0.05) are colored in light 241 grey. 242 243 3.3 Spatial clustering of cocoa, forests and conflict 244 The Hierarchical Clustering Analysis resulted in four clusters of municipalities that are similar with 245 respect to the cocoa production and forest cover variables included in the analysis (Figure 3). The 246 spatial distribution of municipalities defined as ZOMAC and prioritized for PDET show to what 247 degree the municipalities within the four clusters have been affected by the armed conflict and have 248 been prioritized for rural development efforts (Table 2). Figure SM1 in the supplementary material 249 shows the descriptive statistics of the five variables used for the clustering. 250 The results show that cluster 2 and cluster 3 stand out in terms of average forest coverage at the 251 municipal level (45% and 73%, respectively). Furthermore, municipalities located within cluster 2 252 are associated with high levels of forest cover change (on average -0.27%), which is about twice 253 as high as municipalities within the other clusters. Most extensive cocoa production areas are 254 located in municipalities within cluster 3, although the corresponding cocoa yields are relatively 255 low compared to municipalities in other clusters. Municipalities within cluster 4 contain 8 256 considerably more land suitable for cocoa production. Meanwhile, clusters 2 and 3 contain the 257 greatest share of municipalities defined as ZODAC (81; 100% and 54; 86%, respectively) and 258 prioritized for PDET (29; 36% and 41; 65%, respectively). 259 260 Table 2. Characteristics of municipalities within the four clusters, as defined by the Hierarchical Clustering Analysis Cluster 1 Cluster 2 Cluster 3 Cluster 4 (n=192) (n=81) (n=63) (n=193) Average municipal land suitable for cocoa cultivation (%) 19 19 14 57 Average municipal land used to grow cocoa (%) 0.43 0.46 1.68 0.83 Average cocoa yield (ton/ha) 0.58 0.54 0.47 0.58 Average forest area (%) 13 45 73 15 Average rate of forest cover change (%) -0.14 -0.27 -0.15 -0.16 Number of ZOMAC municipalities 83 81 54 78 Number of PDET municipalities 29 29 41 36 261 262 263 Figure 3. Spatial distribution of the clusters of cocoa producing municipalities in Colombia, based on 5 variables related 264 to cocoa production and forest cover. 9 265 266 3.4 Perspective of value chain actors regarding the role of the cocoa sector in delivering forest 267 conservation and peacebuilding 268 Most of the stakeholders interviewed believe that cocoa cultivation has the potential to become a 269 mechanism for peacebuilding and forest conservation – provided that the conditions needed to 270 make cocoa a viable livelihood for families in areas affected by conflict and deforestation are met. 271 Regarding the impact of the cocoa value chain on forest areas, respondents from the public sector 272 assured that forest clearing is inadmissible for cocoa-related development projects. On the other 273 hand, they recognized that plantings have occurred in areas unsuitable for cocoa production. 274 International cooperation actors indicated that they are not carrying out new planting campaigns. 275 Instead, they are focusing on the agricultural intensification of existing plantations, which they 276 claimed does not result in additional deforestation. Most stakeholders acknowledged the potential 277 of cocoa in activities related to reforestation and forest restoration (20 mentions) – particularly in 278 areas that were previously used for the cultivation of coca or converted to pastures for livestock. 279 Three respondents mentioned that the potential for reforestation is evident in areas with illicit crop 280 substitution programs. Meanwhile, 10 respondents mentioned that cocoa-agroforestry systems 281 could positively contribute to biodiversity conservation. Five respondents referred to cases where 282 cocoa production had caused deforestation in the past, but they were uncertain of how much and 283 to what degree deforestation continues to occur. Additionally, one actor pointed out that large-scale 284 industrial plantings pose a threat to forests. Lastly, four respondents mentioned that they could not 285 give an opinion regarding the role of cocoa initiatives in stabilizing the agricultural frontier and 286 halting deforestation because they did not know statistics related to the dynamics of land use 287 changes among forests, degraded pastures and cocoa. 288 Regarding the role of cocoa in peacebuilding and stabilization, 21 actors mentioned that cocoa is a 289 suitable productive alternative for rural areas affected by illicit crops, mainly because it requires 290 similar agro-climatic conditions, enhances community cohesion in the region, encourages 291 associativity and trade networks, raises enthusiasm and commitment for cultivation among the 292 population involved, facilitates learning, and has logistical advantages such as low perishability 293 and greater storability compared to other products. Five actors highlighted the high commercial 294 potential of cocoa (due to favorable prevailing market conditions) as a key advantage of engaging 295 in the cocoa business. 296 Eight stakeholders recognized that the effectiveness of the crop as an instrument for peacebuilding 297 depends on its profitability, which is in turn directly related to the productivity of the system and 298 bean quality. They also mentioned that cocoa cultivation alone is not a solution, but it should be 299 part of a larger bundle of services and investments that are needed in some regions to achieve 300 satisfactory peace outcomes. Four actors mentioned the importance of production diversification, 301 especially in regions affected by the conflict. Three others emphasized the importance of 302 supporting producers during the first 3 to 5 years. During this period, families must assume 303 considerable costs without gaining significant income from cocoa and producers must become 304 familiar with the physiology of the crop. Similarly, two producers stated that technical support 305 alone is not sufficient to turn cocoa production into a viable livelihood in conflict-affected 306 territories. They pointed out that starting cocoa producers are unable to generate income from 307 temporary crops such as plantain during the initial cocoa-growing period, due to poor accessibility 308 of plantain and other crop markets. This poses a considerable challenge for the use of cocoa in 309 promoting peace, as lack of sufficient income to meet the basic needs of producers could reinforce 10 310 the initial causes of the conflict. In this regard, two actors mentioned that there are more suitable 311 alternatives to the widely promoted conventional crop arrangements, such as agroforestry systems 312 of cocoa, bananas and timber. They mentioned that some communities have started implementing 313 systems with lower cocoa densities and greater varieties of species, such as nitrogen-fixing timber, 314 fruit trees, and short-cycle crops that align with the concept of “edible forests,” which could 315 respond better to the needs of some territories. 316 Additionally, actors identified various opportunities to strengthen the current value chain and 317 increase its impact potential. The most commonly identified opportunity comes from the specialty, 318 origin and “fine flavor” cocoa markets, which the actors agree have the competitive advantage. 319 They also recognized other market opportunities, such as cosmetic markets, as easier outlets with 320 capacity to absorb an important share of the national production given their size, lower quality and 321 traceability requirements. On the supply side, the actors emphasized the need for increasing yields 322 through sustainable intensification technologies, which would reduce the amount of land needed 323 to achieve similar output levels. They also emphasized the need for innovation in products and 324 services along the value chain, including sustainable certifications and zero-deforestation 325 commitments (through third party and participative certification approaches), tailor-made financial 326 services, promotion of business services and entrepreneurship along the value chain, and the 327 harnessing of sub-products and byproducts. 328 329 Lastly, actors mentioned several threats and bottlenecks that currently affect the value chain 330 performance – stating that they must be addressed to ensure its sustainability. The main threat is a 331 lack of sufficient volumes with consistent quality and regularity needed to successfully penetrate 332 foreign markets. This is partly attributable to other threats such as low productivity, high 333 transaction costs, insufficient and inefficient institutional support services and lack of capacities 334 along the value chain. Other external factors such as climate change, the EU legislation on 335 cadmium, price variability and improper traceability systems are also recognized as having the 336 potential to negatively impact the value chain and its role in forest conservation and peacebuilding. 337 338 4. Discussion 339 In Colombia, government authorities, international organizations and other stakeholders are 340 looking for opportunities to enhance the performance of agricultural value chains and tackle drivers 341 of deforestation and conflict simultaneously (Castro-Nunez, 2018; Castro-Nunez, Mertz & Sosa, 342 2017). These interventions aim at increasing rural incomes, market access, productivity and welfare 343 to help reconstruct the social fabric and reduce pressure on forests. Most of these programs 344 incorporate environmental components and emphasize reaching international and high-value 345 markets as part of their strategy (Castro-Nunez, 2018). 346 This study explored the potential role of interventions in the cocoa value chain in delivering forest 347 conservation and peacebuilding in Colombia. Findings are consistent with those of previous studies 348 on deforestation in Colombia, in that cocoa is not one of its major drivers (Baptiste et al., 2017; 349 Chadid, Dávalos, Molina & Armenteras, 2015; Dávalos, Holmes, Rodríguez & Armenteras, 2014). 350 National level Spearman’s rank correlation coefficients indicated weak correlations between cocoa 351 production and deforestation, while bivariate Moran’s I values showed limited spatial associations 352 between cocoa production and deforestation at the local level. Together these results suggest that 353 cocoa production is not strongly linked to municipalities that have high rates of forest cover loss. 11 354 Yet in some places, for instance where bivariate Moran’s I values point to High deforestation-High 355 cacao associations, additional analyses at the local level could provide more conclusive evidence 356 of the role cocoa production plays in the deforestation process. 357 The outcomes of the Hierarchical Cluster Analysis show how the degree of cocoa production, 358 forests and conflict varies across Colombian municipalities. This points to biophysical, social and 359 institutional differences among cocoa-producing regions and highlights the importance of adjusting 360 zero-deforestation VCI accordingly. For instance, clusters 2 and 3 contain municipalities with 361 presently the most extensive forest coverage. Hence, focusing zero-deforestation VCI on these 362 municipalities could potentially have a major impact on forest conservation. Even though we find 363 limited evidence that cocoa production drives deforestation, strengthening the cocoa value chain 364 may nevertheless be part of a strategy to reduce deforestation caused by other activities (Castro- 365 Nunez, Bax, Ganzenmuller & Francesconi, 2020). This is particularly true for municipalities 366 subjected to high deforestation rates (i.e. cluster 2), where reinforcing sustainable cocoa production 367 could be a productive alternative to prevailing forest destructive agricultural practices and poor 368 forest management. In a similar manner, in municipalities with high cocoa production levels or 369 extensive areas suitable for production (e.g. cluster 4), incentivizing sustainable agricultural 370 practices for instance through certification and price premiums may be a viable strategy to prevent 371 the expansion of cocoa plantations into forest areas in the future (Castro-Nunez et al.). At the same 372 time, a considerable proportion of municipalities in cluster 2 and cluster 3 have been defined as 373 ZOMAC (100% and 86%, respectively) or prioritized for PDET (36% and 65%, respectively). In 374 consequence, together with the extensive forest coverages within these municipalities, they 375 constitute opportune places to pursue peace and forest conservation objectives simultaneously. 376 Results also show that the majority of stakeholders interviewed report that cocoa is not an important 377 driver of deforestation. Instead, cocoa has the potential to contribute to forest conservation by 378 providing sustainable livelihoods to families involved in economic activities linked to deforestation 379 and address degradation through agroforestry systems. Similarly, they believe that it has a role in 380 peacebuilding as a means for cooperativism, increasing rural incomes and providing licit economic 381 opportunities. Nevertheless, there is a broad agreement that this can only be achieved if the activity 382 becomes a “profitable business”. 383 Results from interviews are consistent with other studies indicating that why and how stakeholders 384 choose to become more sustainable varies along the value chain (Vurro et al., 2009). At the same 385 time, the potential benefits of developing a zero-deforestation cocoa value chain in Colombia are 386 widely recognized. Specifically, there is a strong interest in supporting value chain stakeholders to 387 transition from nationally oriented businesses to engaging in and benefiting from more profitable 388 foreign markets. Interviews suggest that underlying this interest is the assumption that the 389 possibility of getting a better price will incentivize farmers and other stakeholders in the value 390 chain to adopt practices that contribute to improving productivity and comply with social, 391 environmental and quality standards at international levels. Such benefits along the value chain 392 improve livelihood opportunities and could indirectly reinforce public services and institutions that 393 are essential to sustaining peace and achieving the United Nations Sustainable Development Goals 394 (SDG), including deforestation reduction. 395 Findings suggest that efforts to end deforestation in the Colombian cocoa sector emerged following 396 global trends, and not because of an evident link between cocoa production and deforestation. 397 These trends are, nevertheless, consistent with Colombia’s ambitions to strengthen the cocoa 398 sector, mainly in areas affected by the armed conflict, and to do so without harming forests 12 399 (Minambiente, 2018). In this light, the outcomes of this study bring forward a number of key 400 considerations for design and on-the-ground implementation of VCI aimed at preserving tropical 401 forests and cutting carbon emissions linked to the production of globally traded commodities. 402 These considerations are particularly relevant for regions where there is no clear link between 403 agricultural commodity production and deforestation. 404 First, ending deforestation from agricultural commodities will require tailoring global level 405 narratives and approaches to local contexts (Seymour & Harris, 2019). This is particularly true for 406 contexts where the link between agricultural commodities and deforestation is weak, and where 407 most produce is traded nationally and does not involve global brands that dominate the market 408 (Seymour, 2012). For instance, global supply chain interventions as certification programs or 409 moratoria may not be adequate to reach small-scale farmers who produce for domestic or informal 410 markets (Lambin et al., 2018). In addition, global certification programs are less likely to factor in 411 current local production practices, which could open the door to farmers who already comply with 412 production criteria (Blackman & Rivera, 2011) and in turn, reduce the additional forest protection 413 impact of the programs. On the other hand, zero-deforestation initiatives at the national or local 414 level may be more suited to incorporate location-specific problems and development priorities such 415 as peacebuilding, illicit crop eradication, economic growth, rural development, increased 416 agricultural productivity, or increased agricultural exports (Castro-Nunez, Mertz & Quintero, 2016; 417 Castro-Nunez et al., 2017; De Pinto et al., 2016). In the case of Colombia, the goal of achieving 418 zero-deforestation in the cocoa sector as promoted by the Cocoa, Forests and Peace Initiative is 419 compatible with the Colombian government's priorities for reducing coca leaf production and 420 achieving stabilization by developing value chains in conflict-affected areas. In particular, because 421 cocoa production takes place in conflict-affected areas and initiatives emphasize reaching 422 international and high-value markets as a means to increase the profitability of the crop and reduce 423 pressure on forests. 424 425 Second, zero-deforestation commitments, such as those established under the Cocoa, Forests and 426 Peace Initiative in Colombia are an important step toward addressing deforestation from 427 agricultural commodities. Nonetheless, they need to be operationalized by internalizing 428 deforestation concerns in the cocoa value chain. Deans et al. (2018) highlight the importance of 429 strengthening relationships between and collaboration among actors that operate both within and 430 beyond the value chain (e.g. donors, NGO’s and entrepreneurs) to achieve objectives not directly 431 related to production cycle and economic efficiency improvements, such as rural development and 432 deforestation reduction. These advanced collaboration-based governance models focus in part on 433 stimulating the flow of knowledge, finance and information (Bolwig, Ponte, du Toit, Riisgaard & 434 Halberg, 2010) to consolidate the position of smallholders in the value chain and develop a 435 landscape conducive to achieving zero-deforestation outcomes. Some of these concerns also apply 436 to the Colombian context. For instance, our results indicate that farmers may not be able to move 437 from informal business operations and networks to formal and sustainable cocoa production, 438 mainly due to a lack of finance and information. Hence, particularly in the first few years of 439 production, it is fundamental to provide financial support to compensate for low productivity and 440 income losses, and enhance farmers’ capacities related to production practices and 441 entrepreneurship to turn cocoa production into a profitable business. The provisioning of these 442 services in the initial growing-period provide an entry point to zero-deforestation agreements and 443 certified production, wherein — beyond the private sector — a key role is to be played by non- 444 chain actors as NGOs and financial institutions as facilitators of these services. Enhancing this kind 13 445 of stakeholder collaboration within and beyond the value chain could provide the enabling 446 conditions for cocoa production to become viable livelihood, an effective alternative to coca leaf 447 farming, and a tool to prevent future conflict (Nepstad, Boyd, Stickler, Bezerra & Azevedo, 2013). 448 Third, operationalizing zero-deforestation commitments requires big investments. The value chain 449 approach builds on the assumption that companies will not only commit, but actually take 450 ambitious actions to stop deforestation and lower carbon emissions in the tropics. Such actions 451 should be accompanied by supportive public policies to enhance the scale and effectiveness of 452 value chain initiatives and translate them into on-the-ground implementation (Lambin et al., 2018). 453 The current reality is that most rural economies present both a challenging environment for 454 attracting private investment and a difficult arena for public interventions, especially in conflict- 455 affected areas. Thus, zero-deforestation VCI need to be combined with other approaches, engage 456 stakeholders at multiple levels and address multiple factors and interactions to reach zero- 457 deforestation targets (Devaux et al., 2018; Seville et al., 2011). In Colombia, for example, cocoa 458 has been promoted as a productive alternative to illicit crops as part of sustainable rural 459 development strategies. Therefore, it is usually cultivated in areas previously deforested for coca 460 leaf production and where illegal economies predominated. In this context, it is imperative to first 461 support and empower agricultural value chain stakeholders to transform their informal — and 462 sometimes illicit — business activities into formal and professionalized operations that adhere to 463 environmental, social, and quality standards at national and international levels. This will generate 464 an environment conducive for the development of a strong private sector presence, therefore, 465 contributing to long-lasting peace. 466 Fourth, improving our capacity to understand how agricultural commodities are connected to 467 deforestation will help design both zero-deforestation VCI and monitor forest conservation 468 outcomes. For instance, to meet Colombia’s goal of ending deforestation in the Colombian cocoa 469 sector by 2020, the first order of business is to determine where and to what extent deforestation is 470 directly caused by the commodity’s production. Although the results of this study suggest that 471 cocoa was planted in areas previously deforested for other purposes, it may still be the case that 472 cocoa is causing deforestation in some areas of Colombia. Therefore, studies at a lower scale that 473 identify when forest cover changes to a specific crop will help assign causality and attribution, 474 which in turn, forms the basis of increasing traceability and verifying zero-deforestation 475 commitments. This is particularly important if the emphasis on reaching international and high- 476 value markets that value zero-deforestation and peace contributions is part of an incentives 477 strategy. 478 As a final remark, reducing global deforestation may require a transformation of the entire food 479 system. Promoting zero-deforestation in agricultural value chains is undoubtedly a good move. It 480 is not only a way to meet ambitious commitments to preserve tropical forests and cut carbon 481 emissions, but it is also a way to incentivize value chain stakeholders to source, produce, process, 482 and transport agricultural outputs according to environmental, social and quality standards, thus 483 contributing toward sustainable development. Reducing global deforestation, however, will require 484 more than value chain development interventions. It will require changes in farm practices and the 485 farm input supply chain, changes in the intermediating system (change in retail, wholesale, 486 logistics, and processing), and changes on the demand side (diet changes). 487 488 5. Conclusion 14 489 The narrative that agricultural commodities have caused and continue to cause deforestation is 490 starting to dominate the literature and global policies on tropical forest loss. While this is the case 491 for some countries, a different scenario may apply to others. As our study shows, cocoa production 492 has not led to significant deforestation in Colombia. Rather, the government and its development 493 partners are identifying market opportunities to produce cocoa with zero-deforestation in areas 494 prioritized for the peace process. Our study suggests that we need to do further analysis on the links 495 between agricultural commodities and tropical deforestation. This analysis should look into 496 additional scenarios, such as where zero-deforestation VCI have the potential to be used as a tool 497 to overcome barriers to the adoption of sustainable land use systems that contribute toward the 498 restoration of degraded land and prevention of future deforestation (such as agroforestry systems) 499 — particularly in contexts where the link between the commodity and deforestation is not yet 500 evident. 501 Zero-deforestation VCI provide an opportunity to put zero-deforestation commitments by private 502 sector actors into action by creating a framework that facilitates an integrative approach to 503 addressing priorities for economic development and conservation objectives. Promoting and 504 implementing sustainable agricultural value chains in Colombia will require policy coordination 505 across agriculture, forestry and natural resources sectors and the integration of policies that 506 incorporate both conservation goals and the needs of stakeholders. Zero-deforestation VCI, 507 therefore, can be used in Colombia to integrate priorities for agricultural development, 508 environmental conservation and peacebuilding. 15 509 References 510 Abbott, P. C., Benjamin, T. J., Burniske, G. R., Croft, M. M., Fenton, M., Kelly, C. R., et al. (2018). 511 An Analysis of the Supply Chain of cacao in Colombia. United States Agency for International 512 Development (USAID), Cali, Colombia. 513 514 Anselin, L. (1995). Local Indicators of Spatial Association—LISA. Geographical analysis 27(2), 515 93-115. https://doi.org/10.1111/j.1538-4632.1995.tb00338.x. 516 517 Anselin, L., Syabri, I. and Kho, Y. (2006). GeoDa: An Introduction to Spatial Data Analysis. 518 Geographical analysis 38(1), 5-22. https://doi.org/10.1111/j.0016-7363.2005.00671.x. 519 520 Baptiste, B., Pinedo-Vasquez, M., Gutierrez-Velez, V. H., Andrade, G. I., Vieira, P., Estupiñán- 521 Suárez, L. M., et al. (2017). Greening peace in Colombia. Nature Ecology & Evolution 1(4), 0102. 522 https://doi.org/10.1038/s41559-017-0102. 523 524 Blackman, A. and Rivera, J. (2011). Producer‐level benefits of sustainability certification. 525 Conservation Biology 25(6), 1176-1185. https://doi.org/10.1111/j.1523-1739.2011.01774.x. 526 527 Bolwig, S., Ponte, S., du Toit, A., Riisgaard, L. and Halberg, N. (2008). Integrating poverty, gender 528 and environmental concerns into value chain analysis: A conceptual framework and lessons for 529 action research. Danish Institute for International Studies (DIIS), Working paper 2008, 16, 530 Copenhagen, Denmark. 531 532 Bolwig, S., Ponte, S., du Toit, A., Riisgaard, L. and Halberg, N. (2010). Integrating poverty and 533 environmental concerns into value‐chain analysis: a conceptual framework. Development Policy 534 Review 28(2), 173-194. https://doi.org/10.1111/j.1467-7679.2010.00480.x. 535 536 Boucher, D. and Elias, P. (2013). From REDD to deforestation-free supply chains: the persistent 537 problem of leakage and scale. Carbon Management 4, 473-475. https://doi.org/10.4155/cmt.13.47. 538 539 Castro-Nunez, A. (2018). Responding to Climate Change in Tropical Countries Emerging from 540 Armed Conflicts: Harnessing Climate Finance, Peacebuilding, and Sustainable Food. Forests 541 9(10), 621. https://doi.org/10.3390/f9100621. 542 543 Castro-Nunez, A., Bax, V., Ganzenmuller, R. and Francesconi, W. (2020). Emerging scenarios on 544 the role of supply chain initiatives in reducing deforestation: evidence from Peru. Global 545 Environmental Change, (under review). 546 547 Castro-Nunez, A., Mertz, O., Buritica, A., Sosa, C. C. and Lee, S. T. (2017). Land related 548 grievances shape tropical forest-cover in areas affected by armed-conflict. Applied Geography 85, 549 39-50. https://doi.org/10.1016/j.apgeog.2017.05.007. 550 551 Castro-Nunez, A., Mertz, O. and Quintero, M. (2016). Propensity of farmers to conserve forest 552 within REDD+ projects in areas affected by armed-conflict. Forest Policy and Economics 66, 22- 553 30. https://doi.org/10.1016/j.forpol.2016.02.005. 554 555 Castro-Nunez, A., Mertz, O. and Sosa, C. C. (2017). Geographic overlaps between priority areas 556 for forest carbon-storage efforts and those for delivering peacebuilding programs: Implications for 557 policy design. Environmental research letters 12(5), 054014. https://doi.org/10.1088/1748- 558 9326/aa6f20. 559 16 560 Chadid, M. A., Dávalos, L. M., Molina, J. and Armenteras, D. (2015). A Bayesian spatial model 561 highlights distinct dynamics in deforestation from coca and pastures in an Andean biodiversity 562 hotspot. Forests 6(11), 3828-3846. https://doi.org/10.3390/f6113828. 563 564 Charry, A., Castro-Llanos, F. and Castro-Nunez, A. (2019). Colombian cacao, forests and peace 565 initiative = Estudio de línea base de la cadena del cacao en Colombia. International Center for 566 Tropical Agriculture (CIAT), Cali, Colombia. 567 568 Dávalos, L. M., Holmes, J. S., Rodríguez, N. and Armenteras, D. (2014). Demand for beef is 569 unrelated to pasture expansion in northwestern Amazonia. Biological Conservation 170, 64-73. 570 https://doi.org/10.1016/j.biocon.2013.12.018. 571 572 De Pinto, A., Li, M., Haruna, A., Hyman, G. G., Martinez, M. A. L., Creamer, B., et al. (2016). 573 Low emission development strategies in agriculture. An agriculture, forestry, and other land uses 574 (AFOLU) perspective. World Development 87, 180-203. 575 https://doi.org/10.1016/j.worlddev.2016.06.013. 576 577 Deans, H., Ros-Tonen, M. A. and Derkyi, M. (2018). Advanced Value Chain Collaboration in 578 Ghana’s Cocoa Sector: An Entry Point for Integrated Landscape Approaches? Environmental 579 Management 62(1), 143-156. https://doi.org/10.1007/s00267-017-0863-y. 580 581 Devaux, A., Torero, M., Donovan, J. and Horton, D. (2018). Agricultural innovation and inclusive 582 value-chain development: a review. Journal of Agribusiness in Developing and Emerging 583 Economies 8(1), 99-123. https://doi.org/10.1108/JADEE-06-2017-0065. 584 585 Eufemia, L., Bonatti, M., Castro-Nunez, A., Lana, M., Morales, H. and Sieber, S. (2019). 586 Colombia's inadequate environmental goals. Science 364, 444-445. 587 https://doi.org/10.1126/science.aax3494. 588 589 Gardner, T. A., Benzie, M., Börner, J., Dawkins, E., Fick, S., Garrett, R., et al. (2019). 590 Transparency and sustainability in global commodity supply chains. World Development 121, 163- 591 177. https://doi.org/10.1016/j.worlddev.2018.05.025. 592 593 Garrett, R. D., Levy, S., Carlson, K. M., Gardner, T. A., Godar, J., Clapp, J., et al. (2019). Criteria 594 for effective zero-deforestation commitments. Global Environmental Change 54, 135-147. 595 https://doi.org/10.1016/j.gloenvcha.2018.11.003. 596 597 Gibbs, H. K., Rausch, L., Munger, J., Schelly, I., Morton, D. C., Noojipady, P., et al. (2015). 598 Brazil's soy moratorium. Science 347(6220), 377-378. https://doi.org/10.1126/science.aaa0181. 599 600 Henders, S., Persson, U. M. and Kastner, T. (2015). Trading forests: land-use change and carbon 601 emissions embodied in production and exports of forest-risk commodities. Environmental research 602 letters 10(12), 125012. http://doi.org/10.1088/1748-9326/10/12/125012. 603 604 IDEAM (2018). Instituto de Hidrología, Meteorología y Estudios Ambientales. Resultados 605 monitoreo de la deforestación 2017. 606 http://documentacion.ideam.gov.co/openbiblio/bvirtual/023835/Resultados_Monitoreo_Deforesta 607 cion_2017.pdf Accessed 2-12-2019. 608 609 Jäger, M., Jiménez, A. and Amaya, K. (2013). Las cadenas de valor de los ajies nativos de Bolivia: 610 Compilacion de los estudios realizados dentro del marco del proyecto" Rescate y promocion de 611 ajíes nativos en su centro de origen" para Bolivia. Bioversity International, Cali, Colombia. 612 17 613 Jiang, B. (2009). The effects of interorganizational governance on supplier's compliance with SCC: 614 An empirical examination of compliant and non-compliant suppliers. Journal of Operations 615 Management 27(4), 267-280. https://doi.org/10.1016/j.jom.2008.09.005. 616 617 Kidoido, M. and Child, K. (2014). Evaluating value chain interventions: A review of recent 618 evidence. International Livestock Research Institute (ILRI), Discussion paper 26, Nairobi, Kenya. 619 620 Lambin, E. F., Gibbs, H. K., Heilmayr, R., Carlson, K. M., Fleck, L. C., Garrett, R. D., et al. (2018). 621 The role of supply-chain initiatives in reducing deforestation. Nature Climate Change 8(2), 109- 622 116. https://doi.org/10.1038/s41558-017-0061-1. 623 624 MADR (2018). Ministerio de Agricultura y Desarrollo Rural. Portal de Datos Abiertos, Cadena 625 Productiva Cacao -Area, Producción Y Rendimiento. https://www.datos.gov.co/Agricultura-y- 626 Desarrollo-Rural/Cadena-Productiva-Cacao-Area-Producci-n-Y-Rendimie/7rr4-csic/data 627 Accessed 2-12-2019. 628 629 Maestre, M., Poole, N. and Henson, S. (2017). Assessing food value chain pathways, linkages and 630 impacts for better nutrition of vulnerable groups. Food Policy 68, 31-39. 631 https://doi.org/10.1016/j.foodpol.2016.12.007. 632 633 McCarthy, S. and Tacconi, L. (2011). The political economy of tropical deforestation: assessing 634 models and motives. Environmental Politics 20(1), 115-132. 635 https://doi.org/10.1080/09644016.2011.538171. 636 637 Minambiente (2018). Ministerio de Ambiente y Desarrollo Sostenible. Iniciativa Cacao, Bosques 638 y Paz en Colombia Acuerdo Marco para la Acción Conjunta. 639 https://www.minambiente.gov.co/images/BosquesBiodiversidadyServiciosEcosistemicos/pdf/Ac 640 uerdo_cero_deforestacion/cacao.pdf Accessed 2-12-2019. 641 642 Nepstad, D. C., Boyd, W., Stickler, C. M., Bezerra, T. and Azevedo, A. A. (2013). Responding to 643 climate change and the global land crisis: REDD+, market transformation and low-emissions rural 644 development. Philosophical Transactions of the Royal Society B: Biological Sciences 368(1619), 645 20120167. https://doi.org/10.1098/rstb.2012.0167. 646 647 Perez-Aleman, P. and Sandilands, M. (2008). Building value at the top and the bottom of the global 648 supply chain: MNC-NGO partnerships. California management review 51(1), 24-49. 649 https://doi.org/10.2307/41166467. 650 651 R Core Team R: A language and environment for statistical computing. R Foundation for Statistical 652 Computing, Vienna, Austria. 2017. https://www.R-project.org/. 653 654 Reardon, T., Echeverria, R., Berdegué, J., Minten, B., Liverpool-Tasie, S., Tschirley, D., et al. 655 (2019). Rapid transformation of food systems in developing regions: Highlighting the role of 656 agricultural research & innovations. Agricultural systems 172, 47-59. 657 https://doi.org/10.1016/j.agsy.2018.01.022. 658 659 Reed, A. M. and Reed, D. (2009). Partnerships for development: Four models of business 660 involvement. Journal of business ethics 90(1), 3. https://doi.org/10.1007/s10551-008-9913-y. 661 662 Rein, M. and Stott, L. (2009). Working together: Critical perspectives on six cross-sector 663 partnerships in Southern Africa. Journal of business ethics 90(1), 79-89. 664 https://doi.org/10.1007/s10551-008-9915-9. 665 18 666 Schroth, G., Garcia, E., Griscom, B. W., Teixeira, W. G. and Barros, L. P. (2016). Commodity 667 production as restoration driver in the Brazilian Amazon? Pasture re-agro-forestation with cocoa 668 (Theobroma cacao) in southern Pará. Sustainability Science 11(2), 277-293. 669 https://doi.org/10.1007/s11625-015-0330-8. 670 671 Seville, D., Buxton, A. and Vorley, W. (2011). Under what conditions are value chains effective 672 tools for pro-poor development? International Institute for Environment and Development (IIED). 673 674 Seymour, F. (2012). REDD reckoning: a review of research on a rapidly moving target. CAB 675 Reviews: Perspectives in Agriculture Veterinary Science Nutrition and Natural Resources 7(032), 676 1-13. https://doi.org/10.1079/PAVSNNR20127032. 677 678 Seymour, F. and Harris, N. L. (2019). Reducing tropical deforestation. Science 365(6455), 756- 679 757. 680 681 Sola, P., Cerutti, P. O., Zhou, W., Gautier, D., Iiyama, M., Schure, J., et al. (2017). The 682 environmental, socioeconomic, and health impacts of woodfuel value chains in Sub-Saharan 683 Africa: a systematic map. Environmental Evidence 6(1), 4. https://doi.org/10.1186/s13750-017- 684 0082-2. 685 686 Tallontire, A. and Vorley, B. (2005). Achieving fairness in trading between supermarkets and their 687 agrifood supply chains. UK Food Group Briefing. 688 689 Ton, G., Vellema, S. and de Ruyter de Wildt, M. (2011). Development impacts of value chain 690 interventions: how to collect credible evidence and draw valid conclusions in impact evaluations? 691 Journal on chain and network science 11(1), 69-84. https://doi.org/10.3920/JCNS2011.x188. 692 693 Vurro, C., Russo, A. and Perrini, F. (2009). Shaping sustainable value chains: Network 694 determinants of supply chain governance models. Journal of business ethics 90(4), 607-621. 695 https://doi.org/10.1007/s10551-010-0595-x. 696 697 Zuberi, S., Mehmood, R. and Gazdar, H. (2016). Review of Agri-Food Value Chain Interventions 698 Aimed at Enhancing Consumption of Nutritious Food by the Poor: Pakistan. Leveraging 699 Agriculture for Nutrition in South Asia (LANSA), Working paper series, Volume 2016, No7, UK. 19